The satellite technology that detects skin cancer in seconds
“Our mission is to save lives,” says Hans Hekland, project lead of ScreenCancer Mole Spotter (SCMS). His project addresses rising rates of skin cancer in Europe, especially among young people, due to sun exposure and the use of tanning beds. Malignant melanomas are responsible for 7 % of skin cancer deaths, but it can be very difficult to differentiate them from harmless moles or freckles. The only way to secure an early diagnosis of a malignant melanoma is with a specialised tool called a dermatoscope. Yet few physicians have access to one, and those that do may nonetheless pass the results to a trained dermatologist for closer inspection. The difficulty in identifying skin cancers also means that physicians tend to err on the side of caution. As such dermatologists spend much time examining innocent blemishes, and pathologists testing benign tissue. This creates a bottleneck in diagnosing skin cancers. “In the UK, for example, you are guaranteed to see a dermatologist within two weeks where there is a suspicion of cancer,” says Hekland, who is also chairman of ScreenCancer. “The caution of the general practitioners creates a high number of referrals, putting pressure on the healthcare system, leaving other curable skin diseases at six months wait or more.” Through their partnership with pharmacy Boots, ScreenCancer offers a mole checking service in 300 locations across Europe. Dermatoscope images are sent to a qualified dermatologist, with the patient receiving a diagnosis via their mobile phone within a few days. “It’s a good system which worked nicely,” says Hekland. “Over the last years we have investigated more than 100 000 patients and found a high number of malignant melanomas.” However, the process is stymied by a lack of available dermatologists and the time-consuming process of diagnosing each mole. Hekland hopes to solve this with an artificial intelligence that can reduce the time it takes dermatologists to separate malignant melanomas from harmless birthmarks. Working with a specialised camera and optical transfer technology – developed by NASA to monitor the depth of polar ice from orbit – his team developed an algorithm that can analyse images of suspect moles. The artificial intelligence correctly identifies malignant melanomas 99 % of the time. It can also correctly dismiss benign skin blemishes 73 % of the time. The results were published in the British Journal of Dermatology in 2018. “The challenge is that the instrument used for the study was very expensive,” says Hekland. ScreenCancer is now adjusting the algorithm to perform a similar feat on dermatoscopic images. Hekland estimates that the cost of developing this software is around EUR 5 million. “You need many pre biopsy images along with the biopsy result so that the algorithm can learn from it,” he explains. The grant ScreenCancer received will be used to position themselves to apply for this larger funding. Even if his company is successful, humans will remain at the heart of the diagnostic process. “We don’t want to rule out the dermatologist,” says Hekland. “But the number of minutes needed to ensure that deadly cancers are found has to be considerably reduced”.
Keywords
SCMS, ScreenCancer, cancer, melanoma, malignant, mole, skin, freckle, satellite, camera, dermatologist, screen
